The problem isn't really about localization priority of 200Hz vs. 2kHz under free field conditions, it's about what the room-speaker interaction does to the stereo signal around and above the Schröder frequency. The rather large level differences of the steady state response between L and R distort localization cues of the stereo signal.
Hence the real question is probably "can speaker directivity help decreasing localization cue distortion"?
I was thinking about the ITD and IID issues, and it seems that these are parts of the story only. Even when we are not in the position for ITD or IID to theoretically allow identifying sound source location, we often are still able to know where the sound is coming from. I know we can get closer to this kind of presentation because I am beginning to experience it. Although the speaker used does have quite good directivity pattern, it may not be the main reason because I had not experienced it until I got the amplifier to work with the speaker more ideally. Although not as dramatic, I can lay sideways on a couch and still enjoy an opera.
It is critical to have as simplistic wave front as possible with the least drivers as possible.
It is critical to have as simplistic wave front as possible with the least drivers as possible.
Well, the LX should then be more "dynamics capable" than the Orion as it solves the tweeter issue ? A lot of people ( I was one) asked SL about the possibility to use the Seas TDFC27, much cheaper then the Millenium, and still measure admirably. SL did point to possible dynamic issues with it..
Just to get this right for me: You are talking of the "room" response, which is overlayed by the stereo signal? If so, wouldn't one try to have that symmetrical between left and right? By having the stereo triangle axis symmetric to the room axis.
Rudolf
Theoretically placing speakers symmetrically should prevent such errors but in practice the room often imposes very different filter characteristics upon the left and right speaker signal.
There is only heat generation and heat lost to the surroundings. To within some constants, the time rate of change of the temperature is equal to the heat generated in the VC minus the heat lost to the surrounding by any means. So lets look first at the generation part. Obviously, if the heat generation is zero the temperature doesn't change. If the generation term is nonzero then its magnitude determines the rate of change in temperature. Note, it is the rate of change, not the increase. The increase in the integral of the rate on change. If heat generation is constant, in the absence of cooling, the temperature increases linearly over time. The amount of increase is determined by the length of time and the magnitude of the generation. Just like my interest rate example. Also note that it doesn't matter what temperature you start at as long as the material properties are constant. Now, also note that as soon as the generation term goes to zero the rate of change goes to zero and the temperature stops increasing immediately. Now when you add cooling its effect is to damp the system, i.e. slows the response. The VC temp can not increase as fast because some of the generated heat will be rejected to the surroundings.
Anyway, what all this says is that if you are only interested in how fast the VC temp can respond to changes in the current and follow the transient behavior of the signal then all you need to do is consider the heat generation. That sets the upper limit on response time. So basically you end up with
dT/dt = (V(t)^2/R(T)) / (M x Cv)
where dT/dt is the time rate of change of VC temperature, M is the mass of the voice coil, Cv is the heat capacity, V(t) is the voltage across the VC and R(T) is the VC resistance which is a function of temperature. Put in numbers for these quanities and you can calculate dT/dt in degrees/sec which tells you the fastest the termerature can change, temperature slew rate if you like.
In my write up is considered a 20 msec, 100w pulse. This figure shows the part of that signal that would reach the tweeter if the tweeter had a 1k Hz LR2 high pass filter
An externally hosted image should be here but it was not working when we last tested it.
As is seen, the pulse has a sharp rise and then decays in about 1.25 msec.
This next figure shows how the tweeter reacts, not to the actually pulse, but to a constant 100w input. This is much more power than the tweeter would actually receive through the LR2 filter. Never the less, after 1.25msec there is no significant change in sensitivity (no compression).
An externally hosted image should be here but it was not working when we last tested it.
any thermal compression takes place over time due to the more or less time averaged power applied to the driver. This last figure, also from my write up, shows what happens for an actually music signal. In this case the signal is a sample from The Rite of Spring.
An externally hosted image should be here but it was not working when we last tested it.
It represents 70 msec of the signal with many transients and as is plan to see, the sensitivity in no way follows the signal. The thermal response time is just too slow.
Thanks, Markus 
Stereo is interaction between left and right - not so much front/back or top/bottom. Therefore it seems logical that it is more sensitive to left-right asymmetries in the room than other directions. When listening to single sources we must be immune to these room distortions - up to a degree. After all, if someone is walking around the room and speaking in our direction, we don't hear his voice change. Nor does at any time appear his voice to come from another location than his mouth.
Stereo phantoms are of cause less robust in this regard than single sources.
Stereo is interaction between left and right - not so much front/back or top/bottom. Therefore it seems logical that it is more sensitive to left-right asymmetries in the room than other directions. When listening to single sources we must be immune to these room distortions - up to a degree. After all, if someone is walking around the room and speaking in our direction, we don't hear his voice change. Nor does at any time appear his voice to come from another location than his mouth.
Stereo phantoms are of cause less robust in this regard than single sources.
Thanks, Markus
Stereo is interaction between left and right - not so much front/back or top/bottom. Therefore it seems logical that it is more sensitive to left-right asymmetries in the room than other directions. When listening to single sources we must be immune to these room distortions - up to a degree. After all, if someone is walking around the room and speaking in our direction, we don't hear his voice change. Nor does at any time appear his voice to come from another location than his mouth.
Stereo phantoms are of cause less robust in this regard than single sources.
Asymmetric room shape in any dimension will change the balance between left and right. Imagine the wall behind the left speaker is farther away than on the right side. This will cause a level difference between left and right which in turn can cause a change in ITD.
Changing wall properties can act in the same way.
Pipe Organ? Didn't Hermholz use organ pipes for his tests? Of course not all pipes in the organ are that type.
rdf,
I wasn't pointing that comment at you, but in general. It is very difficult to speak of some phenomena when you only look at a discrete frequency band was what I was trying to point at. That is not the real case on musical signal. Conjecture based on sine way analysis and music program material are just not analogous to each other and that was what I was pointing at. So perhaps Earl is correct in the regards of only looking at the limited frequencies in the lower frequency range they would be true, but that isn't the results we get from the complex waveform of any real music accounting for harmonics and overtones in real music.
Rudolph,
I won't argue your mathematics, but this is a real phenomena spoken of in the AES Journal in high power loudspeaker devices running at elevated level over long periods of time. If I had my journal at hand I could point to the technical articles written about this factor. We are both missing something here, but I can't honestly point to the missing factor. But I do hope that you agree that increasing the temperature of a wire does change its electrical resistance, and therefor can change the devices output.
I wasn't pointing that comment at you, but in general. It is very difficult to speak of some phenomena when you only look at a discrete frequency band was what I was trying to point at. That is not the real case on musical signal. Conjecture based on sine way analysis and music program material are just not analogous to each other and that was what I was pointing at. So perhaps Earl is correct in the regards of only looking at the limited frequencies in the lower frequency range they would be true, but that isn't the results we get from the complex waveform of any real music accounting for harmonics and overtones in real music.
Rudolph,
I won't argue your mathematics, but this is a real phenomena spoken of in the AES Journal in high power loudspeaker devices running at elevated level over long periods of time. If I had my journal at hand I could point to the technical articles written about this factor. We are both missing something here, but I can't honestly point to the missing factor. But I do hope that you agree that increasing the temperature of a wire does change its electrical resistance, and therefor can change the devices output.
Pano,
What I am pointing out even with a pipe organ is not in the steady state, but when you first activate the key on the keyboard you have an impulse at the beginning that has a leading edge to it. This is part of the missing equation here, the initialization of the wavefront. When that valve or reed first starts up the ramp to a steady state.
What I am pointing out even with a pipe organ is not in the steady state, but when you first activate the key on the keyboard you have an impulse at the beginning that has a leading edge to it. This is part of the missing equation here, the initialization of the wavefront. When that valve or reed first starts up the ramp to a steady state.
That, John, is good, solid, and "objective" (your whole analysis of the subject is spot-on). As a result it will go . . . nowhere . . . with the people who are convinced that dipoles, ORION in particular, "can't handle dynamics". That's because what they are talking about is not linear amplitude response to the inbut signal. They are talking only about "loud".
What dipoles in general, and ORION in particular can't do is "turn it up to 11". That is what horns do, and do best, and it is their sole advantage. And marketing that as "dynamic", and mistaking that for "dynamic", is what this diversion is all about . . .
Were you adressing ME? Certainly not with an argument I made ...
Would be nice if you could clear that up.
Rudolf
Strictly speaking any additive synthesizer since they generate sounds by adding pure sine wave which are usually harmonically related.
As it goes for true digital synthesizers they are better at creating believable emulations of acoustic instruments than other types like subtractive synths for example.
John K
Nice analysis, but I am still not sure that I agree with your conclusions. First, what size voice coil, and hence mass and thermal capacity are you using? Clearly you can see that these values make a huge difference. Next Why show sensitivity in dB? which would hide any variations of significance. And why just 70 msec? That's pretty short. Why assume that the effect, which clearly increases with time, is only applicable over this very short time frame?
Finally your analysis does not agree with the measurements that I made, which showed thermal changes for a 1" tweeter to be on the order of 2-3 dB, albeit over much longer time scales.
Hence, I am still not convinced that the effect does not exist, although I admit that I am not sure how audible it is. If you plug different values of mass and thermal capacity in your temperature rate equation you will find orders of magnitude difference between a compression driver and a 1" direct radiation tweeter.
If "dynamics" is not thermal then what is it? That question still remains on the table if your conclusion is correct. I have tried to track down this well acknowledged subjective effect and thermal is all that I can come up with.
Finally
Nice analysis, but I am still not sure that I agree with your conclusions. First, what size voice coil, and hence mass and thermal capacity are you using? Clearly you can see that these values make a huge difference. Next Why show sensitivity in dB? which would hide any variations of significance. And why just 70 msec? That's pretty short. Why assume that the effect, which clearly increases with time, is only applicable over this very short time frame?
Finally your analysis does not agree with the measurements that I made, which showed thermal changes for a 1" tweeter to be on the order of 2-3 dB, albeit over much longer time scales.
Hence, I am still not convinced that the effect does not exist, although I admit that I am not sure how audible it is. If you plug different values of mass and thermal capacity in your temperature rate equation you will find orders of magnitude difference between a compression driver and a 1" direct radiation tweeter.
If "dynamics" is not thermal then what is it? That question still remains on the table if your conclusion is correct. I have tried to track down this well acknowledged subjective effect and thermal is all that I can come up with.
Finally
Earl,
Thanks for the rebuttal, I don't have the math but I do believe the effect is real. I was talking about high powered drivers used for PA over long periods of time. I guess these guys have never measured an actual speaker and the temperature rise in not only the voicecoil but the magnetic assembly over time. It becomes a runaway affect as the mass surrounding the voicecoil can not absorb the radiated heat fast enough to control the thermal effects and air is a lousy thermal absorber. I don't have my loudspeaker journal from the AES at hand but I know this subject is covered in one of the technical papers, and those are all vetted papers. I do believe that this effect exists at lower levels but not as a drastic affect like a bass speaker running on the thermal limit with max excursion, but I kept trying to point to the micro effects of heating a voicecoil, the effect is just magnitudes lessor in effect, but it is still there.
Thanks for the rebuttal, I don't have the math but I do believe the effect is real. I was talking about high powered drivers used for PA over long periods of time. I guess these guys have never measured an actual speaker and the temperature rise in not only the voicecoil but the magnetic assembly over time. It becomes a runaway affect as the mass surrounding the voicecoil can not absorb the radiated heat fast enough to control the thermal effects and air is a lousy thermal absorber. I don't have my loudspeaker journal from the AES at hand but I know this subject is covered in one of the technical papers, and those are all vetted papers. I do believe that this effect exists at lower levels but not as a drastic affect like a bass speaker running on the thermal limit with max excursion, but I kept trying to point to the micro effects of heating a voicecoil, the effect is just magnitudes lessor in effect, but it is still there.
Last edited: